Bug 466205

Summary: Request backporting of Apache 2.2.6 internal dummy connection -handling
Product: Red Hat Enterprise Linux 5 Reporter: Jari Turkia <redhat-bugzilla>
Component: httpdAssignee: Joe Orton <jorton>
Status: CLOSED DUPLICATE QA Contact: BaseOS QE <qe-baseos-auto>
Severity: medium Docs Contact:
Priority: medium    
Version: 5.2   
Target Milestone: rc   
Target Release: ---   
Hardware: x86_64   
OS: Linux   
URL: http://wiki.apache.org/httpd/InternalDummyConnection
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2008-10-09 08:21:15 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:

Description Jari Turkia 2008-10-09 06:59:12 UTC
Description of problem:
Under heavy (or moderate) load on a server serving dynamic web pages Apache's internal requests may consume all available Apache resources.

Version-Release number of selected component (if applicable):
httpd-2.2.3-11.el5_1.3


How reproducible:
difficult to reproduce, we ran 8 months without any problems

Steps to Reproduce:
1. set up a web-server with dynamic content
2. put it under stress /w hundreds of child-processes running simultaneously
3. watch legitimate client requests fight with Apache's internal requests
  
Actual results:
Site slowing down considerably. Internal requests stop only after Apache's Timeout-directive (RHEL default 120 seconds). Scoreboard has lot of Apache processes in "Sending Reply" -state. These hang up processes consume other resources too (DB-handles, etc.).

Expected results:
Smoothly running site with fast response time for legitimate users. No system resources wasted on Apache's own clumsiness.

Additional info:
The problem can be defeated using a couple of mod_rewrite -directives separating internal requests from others and telling them to "take a hike".

However, this is a gum/glue/duct tape workaround to save system resources. The problem in Apache does not go away by looking into another direction.

Comment 1 Joe Orton 2008-10-09 08:21:15 UTC
Thanks for the request.  This is already scheduled to be fixed in the next update by switching to use of "OPTIONS *" rather than "GET /" in the dummy request.

*** This bug has been marked as a duplicate of bug 367981 ***